Described is a system for performing probabilistic computations on mobile platform sensor data. The system translates a Bayesian model representing input mobile platform sensor data to a spiking neuronal network unit that implements the Bayesian model. Using the spiking neuronal network unit, conditional probabilities are computed for the input mobile platform sensor data, where the input mobile platform sensor data is a time series of mobile platform error codes encoded as neuronal spikes. The neuronal spikes are decoded and represent a mobile platform failure mode. The system causes the mobile platform to initiate a mitigation action based on the mobile platform failure mode.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A system for performing probabilistic computations on mobile platform sensor data, the system comprising: one or more processors and a non-transitory computer-readable medium having executable instructions encoded thereon such that when executed, the one or more processors perform operations of: translating a Bayesian model representing input mobile platform sensor data to a spiking neuronal network unit that implements the Bayesian model; using the spiking neuronal network unit, computing conditional probabilities for the input mobile platform sensor data, wherein the input mobile platform sensor data is a time series of mobile platform error codes encoded as neuronal spikes, wherein the spiking neuronal network unit comprises a plurality of neurons A, B, C, D, and E having the following dynamics: a tonic input causes neuron A to spike, and neuron A causes neuron B to spike, resulting in an increase to a synaptic weight w, neuron B causes neuron C to spike twice, once through neuron D and once through neuron E, and when neuron C spikes, then a delay τ 1 will cause neuron B to spike before neuron A, resulting in a decrease to the synaptic weight w, wherein the synaptic weight w decreases in proportion to itself; decoding the neuronal spikes, wherein the decoded neuronal spikes represent at least one mobile platform failure mode; and causing the mobile platform to initiate a mitigation action based on the at least one mobile platform failure mode.
This invention relates to a system for performing probabilistic computations on sensor data from mobile platforms, such as vehicles or drones, to detect and mitigate failure modes. The system addresses the challenge of processing time-series sensor data in real-time to identify potential failures and trigger corrective actions. The core innovation involves translating a Bayesian model, which represents the probabilistic relationships in sensor data, into a spiking neuronal network (SNN) that mimics biological neural processes. The SNN processes error codes from mobile platform sensors, encoded as neuronal spikes, to compute conditional probabilities and infer failure modes. The SNN consists of five neurons (A, B, C, D, and E) with specific dynamics. Neuron A spikes in response to a tonic input, triggering neuron B, which increases a synaptic weight (w). Neuron B then causes neuron C to spike twice—once via neuron D and once via neuron E. When neuron C spikes, a delay (τ1) ensures neuron B spikes before neuron A, reducing the synaptic weight (w) proportionally to its current value. The system decodes the neuronal spikes to identify failure modes and initiates mitigation actions, such as alerts or system adjustments, to prevent or address failures. This approach leverages the efficiency of SNNs for real-time probabilistic inference in resource-constrained mobile platforms.
2. The system as set forth in claim 1 , wherein the Bayesian model represents mobile platform error codes, mobile platform failure modes, and connections between the mobile platform error codes and the mobile platform.
A system for predictive maintenance of mobile platforms, such as vehicles or machinery, uses a Bayesian model to analyze error codes and failure modes. The Bayesian model includes representations of mobile platform error codes, failure modes, and the relationships between them. This allows the system to identify patterns and correlations that indicate potential failures before they occur. The model processes real-time or historical data from the mobile platform to predict maintenance needs, reducing downtime and repair costs. By mapping error codes to specific failure modes, the system provides actionable insights for proactive maintenance. The Bayesian approach enables probabilistic reasoning, improving accuracy in failure prediction compared to deterministic methods. The system may integrate with onboard diagnostics or external monitoring systems to collect error data. This predictive capability enhances operational efficiency and safety by addressing issues before they escalate. The model can be updated with new data to refine predictions over time. The system is particularly useful in industries where mobile platform reliability is critical, such as transportation, logistics, and industrial equipment.
3. The system as set forth in claim 1 , wherein in order to compute a conditional probability of two input processes X and Y, inputs to a phasic input and the tonic input of the spiking neuronal network unit are defined, wherein the phasic input comes from a logical conjunction of the two input processes X and Y, and the tonic input comes from the input process X, wherein the tonic input corresponds to P(X), and the phasic input corresponds to P(X,Y), resulting in the synaptic weight w converging to P(X,Y)/P(X)=P(Y|X), where P denotes probability, / denotes division, and | denotes a conditional relationship.
This invention relates to a spiking neuronal network system designed to compute conditional probabilities between two input processes, X and Y. The system addresses the challenge of efficiently estimating conditional probabilities in neural networks, which is crucial for applications in machine learning, signal processing, and cognitive modeling. The system includes a spiking neuronal network unit with two distinct input channels: a phasic input and a tonic input. The phasic input is derived from the logical conjunction of the two input processes, X and Y, representing the joint probability P(X,Y). The tonic input is derived solely from the input process X, representing the marginal probability P(X). By configuring the inputs in this manner, the synaptic weight of the network unit converges to the conditional probability P(Y|X), which is mathematically equivalent to P(X,Y)/P(X). The system leverages the spiking behavior of the neuronal network to dynamically adjust the synaptic weight based on the statistical relationships between the input processes. This approach enables real-time computation of conditional probabilities without explicit probabilistic modeling, making it suitable for hardware implementations and low-power applications. The method ensures accurate probability estimation by directly encoding the joint and marginal probabilities into the network's input structure, allowing the synaptic weight to naturally converge to the desired conditional probability value.
4. The system as set forth in claim 1 , wherein the spiking neuronal network unit is decomposable into a convergence unit CON, which causes the synaptic weight w to exponentially decay to a value, and a multiplier unit M, which acts as a logical conjunction, wherein the convergence unit CON and the multiplier unit M correspond to functions CON: X×Y→Z and M:X×Y→Z, respectively, where Z represents the function output, × denotes paired input arguments, and → denotes mapping from input to output.
This invention relates to a spiking neuronal network system designed to model neural computations with improved efficiency and flexibility. The system addresses the challenge of simulating complex neural behaviors while maintaining computational tractability. A key component is a spiking neuronal network unit that can be decomposed into two distinct functional units: a convergence unit (CON) and a multiplier unit (M). The convergence unit (CON) adjusts synaptic weights (w) by causing them to decay exponentially over time, ensuring that the network adapts dynamically to input signals. The multiplier unit (M) functions as a logical conjunction, combining paired input arguments (X and Y) to produce an output (Z). Both units operate as mathematical mappings, where CON maps inputs X and Y to an output Z, and M similarly maps inputs X and Y to an output Z. This decomposition allows for modular design, enabling efficient implementation and scalability in neural network architectures. The system enhances computational efficiency by separating the decay and logical operations, making it suitable for applications in neuromorphic computing and artificial intelligence.
5. A computer implemented method for performing probabilistic computations on mobile platform sensor data, the method comprising acts of: causing one or more processers to execute instructions encoded on a non-transitory computer-readable medium, such that upon execution, the one or more processors perform operations of: translating a Bayesian model representing input mobile platform sensor data to a spiking neuronal network unit that implements the Bayesian model; using the spiking neuronal network unit, computing conditional probabilities for the input mobile platform sensor data, wherein the input mobile platform sensor data is a time series of vehicle error codes encoded as neuronal spikes, wherein the spiking neuronal network unit comprises a plurality of neurons A, B, C, D, and E having the following dynamics: a tonic input causes neuron A to spike, and neuron A causes neuron B to spike, resulting in an increase to a synaptic weight w, neuron B causes neuron C to spike twice, once through neuron D and once through neuron E, and when neuron C spikes, then a delay τ 1 will cause neuron B to spike before neuron A, resulting in a decrease to the synaptic weight w, wherein the synaptic weight w decreases in proportion to itself; decoding the neuronal spikes, wherein the decoded neuronal spikes represent at least one mobile platform failure mode; and causing the mobile platform to initiate a mitigation action based on the at least one mobile platform failure mode.
This invention relates to probabilistic computations for mobile platform diagnostics using spiking neuronal networks. The problem addressed is the need for efficient, real-time analysis of sensor data from mobile platforms, such as vehicles, to detect and mitigate failure modes. Traditional Bayesian models are computationally intensive for embedded systems, so this method translates a Bayesian model into a spiking neuronal network (SNN) to perform probabilistic computations more efficiently. The SNN processes time-series vehicle error codes encoded as neuronal spikes. The network consists of neurons A, B, C, D, and E with specific dynamics: a tonic input triggers neuron A to spike, which then causes neuron B to spike, increasing a synaptic weight (w). Neuron B then triggers neuron C to spike twice—once via neuron D and once via neuron E. When neuron C spikes, a delay (τ1) causes neuron B to spike before neuron A, decreasing the synaptic weight (w) proportionally to its current value. The decoded neuronal spikes indicate a mobile platform failure mode, prompting the system to initiate mitigation actions. This approach leverages SNNs to perform Bayesian inference efficiently, enabling real-time diagnostics and proactive failure mitigation in resource-constrained mobile platforms.
6. The method as set forth in claim 5 , wherein the Bayesian model represents mobile platform error codes, mobile platform failure modes, and connections between the mobile platform error codes and the mobile platform.
This invention relates to predictive maintenance for mobile platforms, such as vehicles or machinery, using Bayesian models to analyze error codes and failure modes. The system addresses the challenge of detecting and predicting potential failures before they occur, reducing downtime and maintenance costs. The Bayesian model integrates mobile platform error codes, failure modes, and their interconnections to assess system health. By mapping error codes to specific failure modes and understanding their relationships, the model identifies patterns that indicate impending failures. The system processes real-time or historical error data to update the Bayesian model, improving accuracy over time. This approach enables proactive maintenance by flagging high-risk conditions based on probabilistic assessments. The invention enhances reliability by leveraging statistical relationships between error codes and failure modes, allowing for early intervention before critical failures occur. The Bayesian framework provides a flexible and adaptive way to incorporate new data and refine predictions as the mobile platform operates. This method is particularly useful in environments where failure prevention is critical, such as transportation, industrial equipment, or aerospace applications. The system may also include additional features like data preprocessing, anomaly detection, and integration with maintenance scheduling tools to streamline operations.
7. The method as set forth in claim 5 , wherein in order to compute a conditional probability of two input processes X and Y, inputs to a phasic input and the tonic input of the spiking neuronal network unit are defined, wherein the phasic input comes from a logical conjunction of the two input processes X and Y, and the tonic input comes from the input process X, wherein the tonic input corresponds to P(X), and the phasic input corresponds to P(X,Y), resulting in the synaptic weight w converging to P(X,Y)/P(X)=P(Y|X), where P denotes probability, / denotes division, and | denotes a conditional relationship.
This invention relates to computing conditional probabilities in spiking neuronal networks, addressing the challenge of efficiently estimating conditional probabilities between input processes in neuromorphic systems. The method involves defining inputs to a spiking neuronal network unit to compute the conditional probability P(Y|X) of two input processes X and Y. The network unit receives two types of inputs: a phasic input and a tonic input. The phasic input is derived from the logical conjunction of X and Y, representing the joint probability P(X,Y). The tonic input is derived solely from X, representing the marginal probability P(X). The synaptic weight w of the network unit converges to the ratio of the phasic input to the tonic input, which equals the conditional probability P(Y|X). This approach leverages the dynamics of spiking neurons to compute conditional probabilities without explicit probabilistic calculations, enabling efficient and biologically plausible inference in neuromorphic systems. The method is particularly useful in applications requiring real-time probabilistic reasoning, such as sensory processing, decision-making, and adaptive learning in artificial neural networks.
8. The method as set forth in claim 5 , wherein the spiking neuronal network unit is decomposable into a convergence unit CON, which causes the synaptic weight w to exponentially decay to a value, and a multiplier unit M, which acts as a logical conjunction, wherein the convergence unit CON and the multiplier unit M correspond to functions CON: X×Y→Z and M:X×Y→Z, respectively, where Z represents the function output, × denotes paired input arguments, and → denotes mapping from input to output.
This invention relates to spiking neuronal networks, specifically addressing the challenge of efficiently modeling synaptic weight dynamics and logical operations within such networks. The method involves a decomposable spiking neuronal network unit that separates the functions of synaptic weight decay and logical conjunction into distinct components. The unit consists of a convergence unit (CON) and a multiplier unit (M). The convergence unit (CON) is responsible for causing the synaptic weight (w) to exponentially decay to a predetermined value, effectively modeling the temporal decay of synaptic strength. The multiplier unit (M) functions as a logical conjunction, performing a logical AND operation between paired input arguments. Both units operate as mappings from input pairs (X×Y) to an output (Z), where X and Y represent the input arguments and Z is the resulting output. The decomposition allows for modular implementation, improving computational efficiency and flexibility in designing spiking neuronal networks. This approach is particularly useful in applications requiring both dynamic weight adaptation and logical processing within artificial neural networks.
9. A computer program product for performing probabilistic computations on mobile platform sensor data, the computer program product comprising: computer-readable instructions stored on a non-transitory computer-readable medium that are executable by a computer having one or more processors for causing the processor to perform operations of: translating a Bayesian model representing input mobile platform sensor data to a spiking neuronal network unit that implements the Bayesian model; using the spiking neuronal network unit, computing conditional probabilities for the input mobile platform sensor data, wherein the input mobile platform sensor data is a time series of mobile platform error codes encoded as neuronal spikes, wherein the spiking neuronal network unit comprises a plurality of neurons A, B, C, D, and E having the following dynamics: a tonic input causes neuron A to spike, and neuron A causes neuron B to spike, resulting in an increase to a synaptic weight w, neuron B causes neuron C to spike twice, once through neuron D and once through neuron E, and when neuron C spikes, then a delay τ 1 will cause neuron B to spike before neuron A, resulting in a decrease to the synaptic weight w, wherein the synaptic weight w decreases in proportion to itself; decoding the neuronal spikes, wherein the decoded neuronal spikes represent at least one mobile platform failure mode; and causing the mobile platform to initiate a mitigation action based on the at least one mobile platform failure mode.
This invention relates to probabilistic computations on mobile platform sensor data using spiking neuronal networks. The problem addressed is the need for efficient, biologically inspired processing of time-series sensor data to detect and mitigate mobile platform failures. The solution involves translating a Bayesian model into a spiking neuronal network that computes conditional probabilities from encoded sensor data. The system processes mobile platform error codes as neuronal spikes. A spiking neuronal network unit with neurons A, B, C, D, and E performs computations. Neuron A spikes in response to tonic input, triggering neuron B, which increases synaptic weight w. Neuron B then causes neuron C to spike twice—once via neuron D and once via neuron E. When neuron C spikes, a delay τ1 ensures neuron B spikes before neuron A, decreasing synaptic weight w proportionally to its current value. The decoded neuronal spikes identify mobile platform failure modes, triggering mitigation actions. This approach leverages spiking neural networks to model probabilistic relationships in sensor data, enabling real-time failure detection and response in mobile platforms. The system dynamically adjusts synaptic weights to reflect changing conditions, improving accuracy in failure prediction. The method is particularly useful for autonomous systems requiring low-latency, energy-efficient processing of sensor data.
10. The computer program product as set forth in claim 9 , wherein the Bayesian model represents mobile platform error codes, mobile platform failure modes, and connections between the mobile platform error codes and the mobile platform.
This invention relates to a computer program product for analyzing mobile platform failures using a Bayesian model. The system addresses the challenge of diagnosing and predicting failures in mobile platforms, such as vehicles or aircraft, by leveraging error codes and failure modes to improve maintenance and reliability. The Bayesian model is trained to represent mobile platform error codes, failure modes, and the relationships between them. Error codes are specific indicators of system malfunctions, while failure modes describe the underlying causes of these malfunctions. The model maps these error codes to their corresponding failure modes, enabling more accurate diagnostics and root cause analysis. The system processes input data, such as sensor readings or error logs, to identify patterns and correlations between error codes and failure modes. By analyzing these connections, the model can predict potential failures before they occur, allowing for proactive maintenance. The Bayesian approach provides probabilistic reasoning, improving the reliability of failure predictions compared to traditional rule-based systems. This invention enhances maintenance efficiency by reducing downtime and repair costs, while also improving safety by preventing catastrophic failures. The model can be integrated into existing diagnostic systems, providing real-time insights for operators and maintenance teams. The use of Bayesian inference ensures adaptability to new data, allowing the system to evolve with advancements in mobile platform technology.
11. The computer program product as set forth in claim 9 , wherein in order to compute a conditional probability of two input processes X and Y, inputs to a phasic input and the tonic input of the spiking neuronal network unit are defined, wherein the phasic input comes from a logical conjunction of the two input processes X and Y, and the tonic input comes from the input process X, wherein the tonic input corresponds to P(X), and the phasic input corresponds to P(X,Y), resulting in the synaptic weight w converging to P(X,Y)/P(X)=P(Y|X), where P denotes probability, / denotes division, and | denotes a conditional relationship.
This invention relates to computing conditional probabilities in spiking neuronal networks, addressing the challenge of efficiently modeling probabilistic dependencies between input processes. The system uses a spiking neuronal network unit with two distinct inputs: a phasic input and a tonic input. The phasic input receives signals from the logical conjunction of two input processes, X and Y, representing the joint probability P(X,Y). The tonic input receives signals from process X alone, representing the marginal probability P(X). The synaptic weight w of the network unit is adjusted based on these inputs, converging to the ratio P(X,Y)/P(X), which equals the conditional probability P(Y|X). This approach leverages the spiking neuronal network's ability to process temporal and probabilistic information, enabling real-time computation of conditional probabilities without explicit probabilistic calculations. The method is particularly useful in applications requiring dynamic probabilistic inference, such as decision-making systems, pattern recognition, and adaptive control. The system avoids the computational overhead of traditional probabilistic methods by embedding the probability computation directly into the neural network's synaptic plasticity mechanisms.
12. The computer program product as set forth in claim 9 , wherein the spiking neuronal network unit is decomposable into a convergence unit CON, which causes the synaptic weight w to exponentially decay to a value, and a multiplier unit M, which acts as a logical conjunction, wherein the convergence unit CON and the multiplier unit M correspond to functions CON: X×Y→Z and M:X×Y→Z, respectively, where Z represents the function output, × denotes paired input arguments, and → denotes mapping from input to output.
This invention relates to a computer program product implementing a spiking neuronal network unit with a novel decomposition structure. The technology addresses the challenge of efficiently modeling neural dynamics in artificial neural networks, particularly in applications requiring precise control over synaptic weight decay and logical operations. The spiking neuronal network unit is designed to decompose into two distinct functional components: a convergence unit (CON) and a multiplier unit (M). The convergence unit (CON) is responsible for causing the synaptic weight (w) to exponentially decay to a specified value, simulating biological neural processes where synaptic strength diminishes over time. The multiplier unit (M) functions as a logical conjunction, performing a logical AND operation between input signals. Both units operate on paired input arguments (X and Y) and map them to an output (Z), where the functions are defined as CON: X×Y→Z and M: X×Y→Z, respectively. This decomposition allows for modular design and efficient implementation of spiking neural networks, enabling precise control over synaptic dynamics and logical operations within the network. The invention enhances computational efficiency and flexibility in neural network modeling, particularly in applications requiring dynamic weight adjustments and logical signal processing.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
March 6, 2019
March 29, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.